-
Attending to Distinctive Moments: Weakly-Supervised Attention Models for Action Localization in Video
AbstractWe present a method for utilizing weakly supervised data for action localization in videos. We focus on sports video analysis, where videos contain scenes of multiple people. Weak supervision gathered from sports website is provided in the form of an action taking place in a video clip, without specification of the person performing the action. Since many frames of a clip can be ambiguous, a novel temporal attention approach is designed to select the most distinctive frames in which to apply the weak supervision. Empirical results demonstrate that leveraging weak supervision can build upon purely supervised localization methods, and utilizing temporal attention further improves localization accuracy.
Related Material
[pdf][bibtex]@InProceedings{Chen_2017_ICCV,
author = {Chen, Lei and Zhai, Mengyao and Mori, Greg},
title = {Attending to Distinctive Moments: Weakly-Supervised Attention Models for Action Localization in Video},
booktitle = {Proceedings of the IEEE International Conference on Computer Vision (ICCV) Workshops},
month = {Oct},
year = {2017}
}
These ICCV 2017 papers are the Open Access versions, provided by the Computer Vision Foundation.
Except for the watermark, they are identical to the accepted versions; the final published version of the proceedings is available on IEEE Xplore.
Except for the watermark, they are identical to the accepted versions; the final published version of the proceedings is available on IEEE Xplore.
This material is presented to ensure timely dissemination of scholarly and technical work.
Copyright and all rights therein are retained by authors or by other copyright holders.
All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright.